Comunicazione
Multi-messenger radiomics of lung cancer: preliminary results for radiomic features stability among different imaging modalities and extraction software.
Robustelli Test A., Bortolotto C., Pinto A., Brero F., Messana G., Cabini R.F., Postuma I., Stella G.M., Galli G., Mariani M., Figini S., Lascialfari A., Filippi A.R., Bottinelli O.M., Preda L.
Currently, lung Magnetic Resonance Imaging (MRI) is not able to reach Computed Tomography (CT) performances. Being a non-invasive technique, different tools are nowadays employed to improve lung MRI quality. On the other hand, remarkably radiomics has revealed a possibly powerful data analysis method transversal to all Imaging techniques, as it allows to extract quantitative data. In this context, our goals are to investigate the statistical reliability of CT and MRI lung cancer radiomics features and the agreement between two freeware radiomics software, Pyradiomics and LIFEx. This study concerned CT and MRI images from 26 patients, analyzed by using 3 different voxels resampling and 66 selected features. Pyradiomics-LIFEx correlation for each imaging technique, and CT-MRI correlation for each radiomics platform, were investigated. Good/excellent Interclass Correlation Coefficient (ICC) reliability was reached by the following features percentage: MRI nearly 90% and CT 85% considering all the resampling, and less than 15% for both radiomics software. ICC reliability of Pyradiomics and LIFEx features was excellent for at least 45% of features.